The high-performance database archiver for the LHC experiments

Each of the Large Hadron Collider (LHC) experiments will be controlled by a large distributed system built with the Supervisory Control and Data Acquisition (SCADA) tool Prozeßvisualisierungs- und Steuerungsystem (PVSS). There will be in the order of 150 computers and one million input/output parameters per experiment. The values read from the hardware, the alarms generated and the user actions will be archived for the later physics analysis, the operation and the debugging of the control system itself. Although the original PVSS implementation of a database archiver was appropriate for standard industrial use, the performance was not sufficient for the experiments. A collaboration was setup between CERN and ETM, the company that develops PVSS. Changes in the architecture and several optimizations were made and tested in a system of a comparable size to the final ones. As a result, we have been able to improve the performance by more than one order of magnitude, and what is more important, we now have a scalable architecture based on the Oracle clustering technology (Real Application Cluster, RAC). This architecture can deal with the requirements for insertion rate, data querying and manageability of the high volume of data, e.g. an insertion rate of > 150,000 changes/s was achieved with a 6 node RAC cluster.